arXiv subjects
Publications and source records attributed to Matt Groff.
We explore a new form of DFT, which we call the Polynomial Transform. It functions over finite fields, and a size $n$ transform takes $O(n)$ operations. In the multitape Turing machine model, it allows us to multiply two $n$ bit numbers in time $n(k^{\log^*{n}} + \log{p})$, where $k$ is a constant and $\log^*{n}$ is the iterated logarithm. One important consequence is that the Network Coding Conjecture is false.
A New Number Theoretic Transform(NTT), which is a form of FFT, is introduced, that is faster than FFTs. Also, a multiplication algorithm is introduced that uses this to perform integer multiplication faster than O(n log n). It uses preprocessing to achieve an upper bounds of (n log n/(log log n/ log log log n). Also, we explore the possibility of O(n) time multiplication via NTTs that require only O(n) operations, using preprocessing.
The problem of P vs. NP is very serious, and solutions to the problem can help save lives. This article is an attempt at solving the problem using a computer algorithm. It is presented in a fashion that will hopefully allow for easy understanding for many people and scientists from many diverse fields. In technical terms, a novel method for solving k-SAT is explained. This method is primarily based on linear algebra and finite fields. Evidence is given that this method may require rougly O(n^3) time and space for deterministic models. More specifically the algorithm runs in time O(P V(n+V)^2) with mistaking satisfiable Boolean expressions as unsatisfiable with an approximate probablity 1 / \Theta(V(n+V)^2)^P, where n is the number of clauses and V is the number of variables. It's concluded that significant evidence exists that P=NP. There is a forum devoted to this paper at http://482527.ForumRomanum.com. All are invited to correspond here and help with the analysis of the algorithm. Source code for the associated algorithm can be found at https://sourceforge.net/p/la3sat.